Hearing impairment in Dutch patients with connexin 26 (GJB2) and connexin 30 (GJB6) mutations

International Journal of Pediatric Otorhinolaryngology (2005) 69, 165—174

www.elsevier.com/locate/ijporl

Hearing impairment in Dutch patients with connexin 26 (GJB2) and connexin 30 (GJB6) mutations$ Regie Lyn P. Santosa,b, Yurii S. Aulchenkoa, Patrick L.M. Huygenb, Kim P. van der Donkc, Ilse J. de Wijsc, Martijn H. Kempermanb, Ronald J.C. Admiraalb, Hannie Kremerb, Lies H. Hoefslootc, Cor W.R.J. Cremersb,* a

Department of Epidemiology and Biostatistics, Genetic Epidemiology Unit, Erasmus Medical Center Rotterdam, Dr. Molewaterplein 50, 3015 GE Rotterdam, The Netherlands b Department of Otorhinolaryngology, University Medical Centre St. Radboud Nijmegen (UMCN), Radboud West, Philips van Leydenlaan 15, 6500 HB Nijmegen, The Netherlands c Department of Human Genetics, University Medical Centre St. Radboud Nijmegen (UMCN), Radboud West, Philips van Leydenlaan 15, 6500 HB Nijmegen, The Netherlands Received 5 May 2004; accepted 16 August 2004

KEYWORDS Hearing impairment; hearing loss; Connexin 26 (CX26); Connexin 30 (CX30); Gap junction beta-2 (GJB2); Gap junction beta-6 (GJB6)

Summary Objective: Despite the identification of mutations in the connexin 26 (GJB2) gene as the most common cause of recessive nonsyndromic hearing loss, the pattern of hearing impairment with these mutations remains inconsistent. Recently a deletion encompassing the GJB6 gene was identified and hypothesized to also contribute to hearing loss. We hereby describe the hearing impairment in Dutch patients with biallelic connexin 26 (GJB2) and GJB2 + connexin 30 (GJB6) mutations. Methods: The audiograms of patients who were screened for GJB2 and GJB6 mutations were analysed retrospectively. Standard statistical testing was done for symmetry and shape, while repeated measurement analysis was used to assess the relation between mutation and severity. Progression was also studied via linear regression analysis. Results: Of 222 hearing-impaired individuals, 35 exhibited sequence variations; of these 19 had audiograms for study. Hearing loss in patients with biallelic ‘‘radical’’ (i.e. deletions, nonsense and splice site) mutations was significantly worse than in the

$ Note: Nucleotide sequence data reported are available in the GenBank database under the following accession numbers: M869849 for exon 2 of connexin 26; U43932 for exon 1, GJB2; and AJ005585 for connexin 30. * Corresponding author. Tel.: +31 24 3614450; fax: +31 24 3540251. E-mail address: [email protected] (Cor W.R.J. Cremers).

0165-5876/$ — see front matter # 2004 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijporl.2004.08.015

166

R.L.P. Santos et al.

wild type and heterozygotes (SAS proc GENMOD, p = 0.013). The presence of at least one missense mutation in compound heterozygotes tends to lead to better hearing thresholds compared to biallelic radical mutations (p = 0.08). One patient with the [35delG] + [del(GJB6-D13S1830)] genotype was severely impaired. Non-progressive hearing impairment was demonstrated in five 35delG homozygotes in individual longitudinal analyses. However a patient with the [299A > C] + [416G > A] genotype showed significant threshold progression in the lower frequencies. Findings on asymmetry and shape were inconclusive. Conclusions: Our data support the hypothesis that severity is a function of genotype and its effect on the amino acid sequence. A bigger cohort is required to establish nonprogressivity more definitively. # 2004 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The connexin 26 (GJB2) gene is the most studied determinant of recessive nonsyndromic hearing loss. The continued interest in this gene lies in its high carrier frequency (0.5—5.4%) across different ethnic groups [1,2]. There are more than 100 sequence variations known [3] yet one mutation — 35delG — comprises about 70—85% of allelic mutations among Northern Europeans [4—7]. In addition, the gene is small (5.5 kb long) and only one of two exons has the coding sequence. All these favor the development of an effective genetic screening method for recessive deafness that is expected to have a high impact on clinical practice and genetic counselling. Yet a phenotypic picture for GJB2 remains incomplete, which makes genetic counselling difficult. The presence of two GJB2 mutations (that is, homozygous or compound heterozygous; from hereon referred to as biallelic) has been associated with hearing impairment that affects all frequencies and that is prelingual, symmetric, sensorineural, non-progressive, and of variable severity even within sibships [4,5,8,9]. The first four characteristics seem consistent across populations. Progression has been described in some research, including some description of sudden progressive hearing loss, but was tested for statistical significance in only a few studies [6,10]. Severity has been extensively discussed and there is evidence that it varies as a function of genotype. Genetic screening was previously limited to 35delG until more studies showed varying patterns of severity in compound heterozygotes and non-35delG homozygotes [6,11,12]. The trend now is to sequence the coding exon and the splice sites of the gene, so that different combinations of mutations may aid in predicting the severity of hearing impairment. In a relatively large study, the hearing loss in 35delG homozygotes still ranged widely from moderate to profound [10], which may be due to yet undis-

covered gene—gene or gene—environment interactions. A deletion in the connexin 30 (GJB6) gene can also cause hearing impairment, either as a homozygous deletion or in combination with a GJB2 mutation [13—16]. Several genetic centers have retrospectively analysed the DNA of proven GJB2 heterozygotes for del(GJB6-D13S1830), with the hope that the variability of hearing loss in GJB2 heterozygotes may be explained by interaction between these genes. The prevailing hypotheses are either: (1) that the connexin genes have a digenic mode of inheritance, so that a dosage effect accounts for the inner ear dysfunction; and/or (2) that the deletion may include a regulatory segment for GJB2, thus inhibiting normal gene expression [13—16]. However there is very little known about the phenotype. We retrieved the audiometric records of hearingimpaired patients who were referred from different regions of The Netherlands to the University Medical Centre St. Radboud Nijmegen (UMCN) for GJB2 and GJB6 screening. Our aim was to describe the hearing impairment, particularly the severity and the progression, in Dutch children that have biallelic GJB2 or GJB6 mutations. Furthermore severity was analysed as a function of both the genotype and the type of mutation according to the change in amino acid sequence.

2. Methods From November 1998 to January 2003, 264 individuals with apparently recessive hearing impairment were referred from various genetic and otolaryngologic clinics in The Netherlands to the UMCN Human Genetics Department for GJB2 and GJB6 testing. Part of this sample (11 patients) was described previously [17,18]. In the analysis we included patients who met the following criteria: (1) no evidence of other causes of genetic hearing loss,

Hearing impairment in Dutch patients

syndromic or nonsyndromic; (2) Dutch Caucasian origin; and (3) available audiometric records from below 18 years of age and onwards. For multiaffected families, only one hearing-impaired individual (proband) was sampled. Interestingly, there were patients with other genetic diseases who also carried connexin 26 mutations: a patient with hereditary neuropathy and polymorphisms [79G > A] + [341A > G]; a child with a 416G > A allele and DFNA3; and, a 35delG carrier with Usher2A syndrome. Another excluded patient was a member of a previously reported family with both GJB2 and GJB6 mutations segregating [19]. Eight adult patients with mutations, including two who had [35delG] + [del(GJB6-D13S1830)] and [167delT] + [del(GJB6-D13S1830)], were not included in the sample for analysis. Genomic DNA was isolated from peripheral blood using standard salt extraction methods. For sequencing of the entire GJB2 coding region, exon 2 was amplified with the primers 50 -CCTATGACAAACTAAGTTGGTTC-30 and 50 -GCCTCATCCCTCTCATGCTGTC-30 . Sequence analysis was performed with amplification primers and additional internal primers (forward primers 50 -GGGGAGATGAGCAGGCCGAC-30 and 50 -CGGCTGGTGAAGTGCAACGC-30 ; reverse primers 50 -CGTAGCACACGTTCTTGCAGCCTG—30 and 50 -CGATGCGGACCTTCTGGGTTTTG30 ) using the ABI PRISM Big Dye Terminator cycle sequencing V2.0 ready reaction kit and ABI PRISM 3700, or 3730 DNA analyser (Applied Biosystems Inc.). For the analysis of the splice donor site of exon 1, primers 50 -TCCGTAACTTTCCCAGTCTCCGAGGGAAGAGG-30 and 50 -CCCAAGGACGTGTGTTGGTCCAGCCCC-30 were used for amplification and sequence analysis. The presence of the deletion encompassing part of the GJB6 gene was analysed essentially as described by del Castillo et al. [14]. In brief, a multiplex PCR that detects both the wild type and the mutant allele was employed. The nomenclature system that was proposed by den Dunnen and Antonarakis was used to denote all sequence variations [20]. When available, documentation of medical history, physical examination and additional laboratory and audiovestibular tests were reviewed. For UMCN patients (n = 106), questionnaires were administered to further screen for other causes of hearing loss, such as recurrent middle ear infections, prolonged use of antibiotics/medication, noise exposure, problems surrounding delivery at birth, head trauma and meningitis. Parents’ consent was obtained in all pediatric cases; the rest of the patients also gave written informed consent.

167

Only pure tone audiometric results that were acquired through standard methods (soundproof booth with a calibrated machine, using ISO standards) were used. For cross-sectional analysis, the latest audiogram was employed. Air-bone gap was checked to rule out middle ear problems at the time of testing. Asymmetry was defined as >10 dB difference in at least two frequencies between the two ears [21]. To measure hearing loss severity, the air conduction pure tone average (PTA) of both ears was computed using three frequencies (500, 1000, and 2000 Hz). Out-of-scale measurements were treated as missing values. Hearing loss was then categorized according to the guidelines of the European Work Group on Genetics of Hearing Impairment: 21—40 dB as mild, 41—70 dB as moderate, 71—95 dB as severe, and more than 95 dB as profound. The criteria set by Liu and Xu were followed for audiogram shape [22]. Using SPSS 11.0.1 (SPSS Inc., 1989—2001) software, Fisher’s exact test, tests between means and nonparametric tests were utilized to check relationships between variables as appropriate. p-Values of less than or equal to 0.05 were considered significant. Because an audiogram measures thresholds for several frequencies (usually four to six) per ear, each test per frequency may be treated as a repeated measurement. Moreover, a patient may have several audiograms available. To avoid multiple repeated measures, only the last audiometric test was analysed using the SAS/STAT Version 8 procedure GENMOD (SAS Institute Inc., Cary, NC USA, 1999) in order to establish the relation between hearing threshold, mutation, gender and age at audiometry. GENMOD applies Generalized Estimation Equations that are less sensitive to violations of the assumption of normality. An autoregressive covariance structure was assumed. The [35delG] + [del(GJB6-D13S1830)] genotype was excluded from this analysis. For longitudinal analysis of progression, linear regression analysis was done for patients with at least three consecutive measurements using GraphPad Prism version 3 software (GraphPad Software Inc., San Diego, CA, USA, 2002). This method of analysis was outlined previously [23].

3. Results Out of 264 referred individuals, 222 were of Dutch descent. Of these 35 patients with no other identifiable cause of genetic hearing impairment screened positive for connexin 26 mutations and were pediatric and/or had audiometric measurements in childhood. Ten were heterozygous for the following

168

R.L.P. Santos et al.

polymorphisms, with frequencies indicated in parentheses if occurring more than once: 35delG (2); 101T > C (3); 42C > G; 249C > G; 355G > A; 457G > A; and 478G > A. There were 1335delG homozygotes and 8 compound heterozygotes ([35delG] + [IVS1 + 1G > A] (2); [35delG] + [109G > A]; [35delG] + [229T > C]; [35delG] + [313_326del]; [35delG] + [449delT]; [101T > C] + [427C > T]; [299A > C]+[416G > A]). Four had connexin 30 plus connexin 26 mutations ([35delG] + [del(GJB6-D13S1830)] (3); [313_326del] + [del(GJB6-D13S1830)]). Presence of mutations was equally distributed between genders. Of the 35 patients with mutations, we were able to retrieve standard audiometric records for 19 (Table 1), 12 of whom were from the UMCN. Twenty-two more Dutch patients from the UMCN were negative for GJB2 or GJB6 mutations and other deafness-causing mutations screened but had pediatric audiograms available. Table 2 shows the mean thresholds for each genotype. For the patients with GJB2 mutations, the average age at first audiometry was 4.4  0.59 years (s.d. = 2.58). For the last audiometry the average age was 12.3  1.95 (s.d. = 8.51). The proportion of cases that had other siblings affected comprised

57.9% (11/19) of those with mutations and audiometry. Most of the mutations (29 of 35 alleles) were expected to change the protein sequence radically; these included deletions, nonsense and splice site mutations. Only six alleles had missense changes. Five of nine 35delG homozygotes demonstrated asymmetric hearing impairment. The patient with the 109G > A mutation and two heterozygotes also showed asymmetric hearing loss. The audiometric shape was highly variable among the biallelic mutations and even between ears, but was usually residual, flat or gently sloping (Table 1). Two 35delG homozygotes had U-shaped curves in one ear. Specifically the [35delG] + [109G > A] genotype resulted in sharply sloping audiograms for both ears. Ascending shapes were found only in heterozygotes. One [35delG] + [del(GJB6-D13S1830)] patient had asymmetric hearing loss: moderate in one ear and severe in the other, with a mean PTA of 81.7 dB. Moreover, the audiogram shape was flat on the right and ascending on the left. The audiologic records of two [35delG] + [del(GJB6-D13S1830)] individuals were not available. The child with the [313_326del] + [del(GJB6-D13S1830)] mutations had profound hearing loss as shown by auditory brainstem response (ABR).

Table 1 Severity, shape and progression in patients with mutations Mutationa

Protein changeb

Severity

Audiometric shape

A. Biallelic mutations [35delG] + [35delG] [35delG] + [35delG]d [35delG] + [35delG] [35delG] + [35delG] [35delG] + [35delG] [35delG] + [35delG] [35delG] + [35delG] [35delG] + [35delG] [35delG] + [35delG] [35delG] + [IVS1 + 1G > A] [35delG] + [IVS1 + 1G > A] [35delG] + [313_326del] [35delG] + [109G > A] [101T > C] + [427C > T] [299A > C] + [416G > A] [35delG] + [del(GJB6-D13S1830)]

R+R R+R R+R R+R R+R R+R R+R R+R R+R R+R R+R R+R R+M M+M M+M R+R

Severec Profoundc Profoundc Profound Profound Profound Profound Profound Severe Profoundc Severe Profound Moderate Moderate Profounde Severe

Gently sloping/U-shaped Residual Flat/gently sloping Residual Residual Residual U-shaped/flat Gently sloping/residual Gently sloping Residual Gently sloping/flat Residual Sharply sloping Flat Residual Flat/ascending

B. Heterozygous mutations [35delG] + Wt [457G > A] + Wt [478G > A] + Wt

R + Wt M + Wt M + Wt

Severe Moderate Moderate

Sharply sloping/gently sloping Ascending Ascending

a

Wt, wild type. R, mutation with radical change in protein sequence; M, missense mutation. c Non-progressive (Fig. 1). d Patient with monozygotic twin. Expected amino acid changes with substitutions: 109G > A = V37I; 101T > C = M34T; 427C > T = R143W; 299A > C = H100P (novel); 416G > A = S139N; 457G > A = V153I; 478G > A = G160S. The last two were identified as polymorphisms, while 101T > C remains controversial. e Progressive (Fig. 2). b

Hearing impairment in Dutch patients

169

Table 2 Severity according to genotype Genotypea

N

Mean PTA (dB)b

Severity

[35delG] + [35delG] [35delG] + [IVS1 + 1G > A] [35delG] + [313_326del] [35delG] + [del(GJB6-D13S1830)] [35delG] + [109G > A] [101T > C] + [427C > T] [299A > C] + [416G > A] [35delG] + Wt [457G > A] + Wt [478G > A] + Wt Wt

9 2 1 1 1 1 1 1 1 1 22

101.5 (s.d. = 11.6) 97.9 106.7 81.7 45.0 40.8 107.5 82.5 70.0 60.8 91.8 (s.d. = 17.1)

Severe to profound Severe to profound Profound Severe Moderate Moderate Profound Severe Moderate Moderate Moderate to profound

a b

Wt, wild type. Values computed from air conduction pure tone averages (0.5—2 kHz) of both ears from latest audiogram per individual.

Twenty-two UMCN patients who were negative for GJB2 and GJB6 and whose audiograms were also available were used as a comparison group to test if a specific picture of the GJB2 phenotype emerges. Age, gender and familiality of hearing loss were not significantly different between the mutation and comparison groups. Differences in shape, air-bone gap and symmetry between the two groups were also not significant. Repeated measures analysis on patients with and without GJB2 mutations showed a significant effect of the mutations, but not age and gender, on the hearing thresholds (SAS proc GENMOD, Wald type 3 statistic = 478.3, p < 0.0001). Based on the parameter estimates, the severity of hearing impairment in subjects with the [35delG] + [35delG], [35delG] + [IVS1+1G > A] and [35delG] + Wt genotypes were not significantly different from the wild type, but those with [35delG] + [109G > A], [101T > C] + [427C > T] and polymorphisms had significantly lower thresholds (data not shown). On the Table 3

other hand, threshold estimates for [35delG] + [313—326del] and [299A > C] + [416G > A] were significantly higher than for wild type. When the mutations were recategorized into 35delG and non-35delG genotypes, these differences became non-significant. However when the grouping was changed according to amino acid sequence variation, i.e. ‘‘radical’’ [24] and missense, different combinations of mutations have a significant effect on the thresholds (SAS proc GENMOD Wald type 3 statistic = 535.6, p < 0.0001, Model A in Table 3). After further reclassification into three groups (i.e. (1) radical + radical mutations, (2) at least one missense mutation, biallelic, (3) wild type and heterozygotes), the association between mutation and hearing threshold remained significant (Wald type 3 statistic = 6.9, p = 0.03, Model B in Table 3). A combination of radical mutations would be significantly worse than wild type and heterozygotes (p = 0.01). The presence of at least one missense mutation also showed a marked improvement in

Generalized estimation equations (GEE) parameter estimates for mutations grouped according to functiona

Model A Radical + radical Radical + missense Missense + missense Radical + Wt Missense + Wt Wild type

Estimate + Se

Z-value

p-Value

Reference 37.9  5.06 23.6  22.55 8.9  5.12 34.9  3.86 8.7  4.63

7.5 1.0 1.7 9.1 1.9

<0.0001 0.3 0.08 <0.0001 0.06

Type 3 Wald: bmutation, d.f. = 5; Chi-square statistic = 535.6; p-value < 0.0001; bage and bgender not significant Model B Radical + radical Missense + missense or radical Wt and mutation + Wt

Reference 29.2  16.69 12.1  4.86

1.8 2.5

Type 3 Wald: bmutation, d.f. = 2; Chi-square statistic = 6.9; p-value = 0.03; bage and bgender not significant a

Regression equation: db  intercept + bmutation  mutation + bage  age + bgender  gender.

0.08 0.01

170

R.L.P. Santos et al.

Table 4 Pooled data on severity according to genotype and amino acid changea N

Severity, n (%)

A. Genotype [35delG] + [35delG] [35delG] + non-35delG Non-35delG + non-35delG [35delG] + Wt Non-35delG + Wt

70 9 2 30 4

45 (64.3%) profound [moderate to profound] 6 (66.7%) profound [moderate to profound] Moderate—profound 15 (50%) profound [mild to profound] Mild—severe

B. Amino acid change Radical + radical Radical + missense Missense + missense Radical + Wt Missense + Wt

78 1 2 30 4

51 (65.4%) profound [moderate to profound] Moderate Moderate—profound 15 (50%) profound [mild to profound] Mild—severe

b

a b

Additional data from references [5,6] and Zoll et al. (2002). Wt, wild type.

threshold; however the analysis showed a p-value of 0.08. We pooled the data on severity of hearing impairment with three other studies from Northern Europe that used the same categorization [5—7]. Only the genotypes similar to those in this study were included. With a sample of 115, a significant association between severity and genotype was seen (Fisher’s exact test, p = 0.009). Grouping into 35delG and non-35delG genotypes also resulted in significant association with severity (Fisher’s, p = 0.006, Table 4), while grouping according to amino acid sequence change was highly significant when tested versus severity (Fisher’s, p < 0.003). It should be noted, however, that this significance is mainly derived from the comparison between 35delG homozygotes and 35delG heterozygotes. Progression was examined in five 35delG homozygotes who had 7—22.7 years of follow-up (mean follow-up = 15.2 years), and was found to be not significant. Two of the patients are identical twins. When data from all five patients were put together, the age range tested was from 2.7 to 30 years old. Fig. 1 shows the plot of one of the twins. For the five 35delG homozygotes, hearing impairment was found to be non-progressive. However the patient with the [299A > C] + [416G > A] genotype had significant progression at the low frequencies (2.2 dB/year at 0.25 kHz, 0.9 dB/year at 0.5 and 2.2 dB/year at 1 kHz, see Fig. 2).

4. Discussion In summary, we did not detect progression of hearing impairment with 35delG homozygotes. We also support the results of previous research that severity is a function of genotype and its effect on the

amino acid sequence. As opposed to the severe or profound hearing loss that was found with deletions and splice site mutations, the genotypes with missense mutations tend to be moderate in hearing impairment. From our sample of hearing-impaired individuals 15.8% (35/222) were positive for mutations, while 37 (66.1%) of 56 mutated GJB2 alleles were 35delG. These percentages are consistent with literature [4—6,8—10,25]. Our average age at first audiometry may seem late for the diagnosis of GJB2 deafness, which is usually prelingual. However earlier measurements by free field and/or observation audiometry were excluded from analysis. We found patients with syndromic or nonsyndromic hearing loss combined with GJB2 mutations. Extra-auditory signs previously reported in GJB2 patients were speculated to be entirely due to GJB2 gene dysfunction [26]. Occurrence of other genetic diseases should always be considered in these cases that might warrant further screening. This also paves the way for the study of yet undiscovered gene—gene interactions. Surprisingly six patients with biallelic mutations had asymmetric hearing loss based on Mazzoli et al.’s criteria of >10 dB difference in 2 frequencies [21]. Interaural differences in GJB2 patients were claimed to be rare [4], and no difference in symmetry was reported between those who screened positive and negative for GJB2 mutations [5]. When we redefined asymmetry as >15 dB difference in at least two frequencies, in only two patients does the hearing loss remain asymmetric. Further use of Mazzoli et al.’s criteria for asymmetry should demonstrate its applicability to datasets. Our results on audiometric curve shape agree with past research that there is no pathognomonic

Hearing impairment in Dutch patients

171

Fig. 1 Serial threshold measurements in a homozygous 35delG patient not showing any significant progression. Air conduction threshold (circles, right; crosses, left) is plotted against age with connecting hairlines and regression lines (solid lines, right; dashed lines, left). Bottom panels show pure tone average (PTA) for each side and binaural PTA (asterisks and dotted lines).

pattern for GJB2, and that flat and sloping curves are common [4,9,27]. However, residual hearing occurred more often in our sample. Also we found U-shaped curves, although only in one ear; ascend-

ing shapes were seen only in heterozygotes. Our data on progression do not support the possibility of temporal changes in shape as previously proposed [27]. Better ascertainment of patients with milder

172

R.L.P. Santos et al.

Fig. 2 Measurable air conduction thresholds plotted against age in a patient with the [299A > C] + [416G > A] genotype. Same symbols as in Fig. 1. Bold regression lines indicate significant progression. Progression was 2.2 dB/ year in both ears at 0.25 kHz, 1.0 dB/year in the right ear and 0.8 dB/year in the left ear at 0.5 kHz, and 2.2 dB/year in the left ear at 1 kHz. Threshold data for the other frequencies and PTA data are not shown because they covered too many out-of-scale measurements to allow for reliable analysis.

hearing impairment may provide more clues concerning shape in the GJB2 genotype. Certain genotypes, specifically the [35delG] + [109G > A] and [101T > C] + [427C > T], had milder phenotypes, as compared to [35delG] + [35delG], [35delG] + [IVS1 + 1G > A] and [35delG] + [313—326del]. This was further confirmed by repeated measurements analysis. Recent research put forward the idea that the wide variability in severity for GJB2 mutations is dependent on the genotype [6,10]. Looking into the effect of these genotypes on the amino acid sequence strengthens this hypothesis even more. The phenotype for the 35delG homozygote remained variable, as in past research, which ranged from moderate to profound. Compared to Cryns et al however, our [35delG] + [109G > A] and [35delG] + [IVS1 + 1G > A] patients exhibited more severe hearing impairment [10]. Cryns et al. also did not include the 101T > C alleles in their analysis. Though much controversy has surrounded this mutation, several studies suggested that the 101T > C is a recessive mutation that may cause milder forms of hearing loss [7,11,12,28,29]. The genotype [299A > C] + [416G > A], though expected to produce a missense change, produced hearing impairment which is worse than in 35delG

homozygotes and with progression at the lower frequencies (Fig. 2). We did not encounter previous reports of this genotype. The affected patient’s younger sibling with the same genotype has severe hearing impairment based on click brainstem audiometry. More individuals may be needed to study the phenotype for this specific genotype. In a multicenter study, heterozygous GJB6 mutations were found in 0-29 patients among different populations and even rarer, del(GJB6-D13S1830) homozygotes in <0.5% [30]. We found 6 del(GJB6D13S1830) heterozygotes in 222 (2.7%) hearingimpaired individuals, which further testifies that this mutation is not as rare as previously thought. To identify the GJB6 deletion, the haplotypes of prelingually and profoundly deaf patients were analysed in previous reports [13,15]. In a study of several multiaffected families from West Flanders, the [35delG] + [del(GJB6-D13S1830)] genotype was also shown to cause profound deafness [19]. Unfortunately, we had only one patient with available audiometry, whose hearing loss was severe rather than profound. More research, especially in populations, is required to confirm these findings. In past research GJB2 mutations were shown to be non-progressive [4,5,9,10]. Although 88% of their sample was non-progressive, Janecke et al.

Hearing impairment in Dutch patients

reported progressivity of hearing loss associated with sudden sensorineural hearing loss in three patients [6]. We demonstrated non-progressive hearing impairment for 35delG homozygotes at age 21/2 and above. Our data on progression agrees with the hypothesis that hearing loss in most GJB2 mutations is prelingual [4]. In studies in mice, hearing impairment began at about the onset of inner ear function [31], which corresponds to the 20th week of gestation in humans. The hair cell damage and death that ensued was attributed to low endolymphatic K+ potentials. The question as to whether hearing impairment progresses prenatally or postnatally may be important, because intervention prior to complete degeneration of the inner ear might result in better habilitation [31]. Since our number of cases is small, we hope that our data can be part of a bigger sample to establish non-progression in connexin 26 mutations. To conclude, we presented additional evidence that GJB2 mutations cause hearing impairment that in general does not progress, and that varies in severity depending on the genotype and its effect on the amino acid sequence. GJB6 mutations were relatively frequent, and may cause a severe phenotype. Further use of new criteria for symmetry and more data on audiogram shape may be required to better establish the genotype—phenotype correlation for the connexin genes.

Acknowledgements The authors would like to thank the participating patients and families and the doctors who provided clinical information. We also thank Prof. Theo Stijnen for statistical advice, and Prof. C.M. van Duijn for general support. This study was sponsored in part by the Heinsius Houboult Foundation (to C.C.) and The Netherlands organization for international cooperation in higher education (Nuffic) (to R.S.).

References [1] R. Rabionet, P. Gasparini, X. Estivill, Molecular genetics of hearing impairment due to mutations in gap junction genes encoding beta connexins, Hum. Mutat. 16 (2000) 190—202. [2] L. Van Laer, P. Coucke, R.F. Mueller, G. Caethoven, K. Flothmann, S.D. Prasad, et al. A common founder for the 35delG GJB2 gene mutation in connexin 26 hearing impairment, J. Med. Genet. 38 (2001) 515—518.

173

[3] J. Calvo, R. Rabionet, P. Gasparini, X. Estivill, Connexins and deafness homepage, World wide web URL: http:// www.crg.es/deafness, Accessed 24 March 2004. [4] F. Denoyelle, S. Marlin, D. Weil, L. Moatti, P. Chauvin, E.N. Garabedian, et al. Clinical features of the prevalent form of childhood deafness, DFNB1, due to a connexin-26 gene defect: implications for genetic counselling, Lancet 353 (1999) 1298—1303. [5] R.F. Mueller, A. Nehammer, A. Middleton, M. Houseman, G.R. Taylor, M. Bitner-Glindzicz, et al. Congenital non-syndromal sensorineural hearing impairment due to connexin 26 gene mutations–—molecular and audiological findings, Int. J. Pediatr. Otorhinolaryngol. 50 (1999) 3—13. [6] A.R. Janecke, A. Hirst-Stadlmann, B. Gunther, B. Utermann, T. Muller, J. Loffler, et al. Progressive hearing loss, and recurrent sudden sensorineural hearing loss associated with GJB2 mutations–—phenotypic spectrum and frequencies of GJB2 mutations in Austria, Hum. Genet. 111 (2002) 145— 153. [7] B. Zoll, L. Petersen, K. Lange, P. Gabriel, C. Kiese-Himmel, P. Rausch, et al. Evaluation of CX26/GJB2 in German hearing impaired persons: mutation spectrum and detection of disequilibrium between M34T (c.101T>C) and -493del10, Hum. Mutat. 21 (2003) 98. [8] E.S. Cohn, P.M. Kelley, T.W. Fowler, M.P. Gorga, D.M. Lefkowitz, H.J. Kuehn, et al. Clinical studies of families with hearing loss attributable to mutations in the connexin 26 gene (GJB2/DFNB1), Pediatrics 103 (1999) 546—550. [9] A. Murgia, E. Orzan, R. Polli, M. Martella, C. Vinanzi, E. Leonardi, et al. CX26 deafness: mutation analysis and clinical variability, J. Med. Genet. 36 (1999) 829—832. [10] K. Cryns, E. Orzan, A. Murgia, Huygen PLM, F. Moreno, I. del Castillo, et al. A genotype—phenotype correlation for GJB2 (connexin 26) deafness, J. Med. Genet. 41 (2004) 147—154. [11] R.A. Cucci, S. Prasad, P.M. Kelley, G.E. Green, K. Storm, S. Willocx, et al. The M34T allele variant of connexin 26, Genet. Test. 4 (2000) 335—344. [12] M.J. Houseman, L.A. Ellis, A. Pagnamenta, W. Di, S. Rickard, A.H. Osborn, et al. Genetic analysis of the connexin-26 M34T variant: identification of genotype M34T/M34T segregating with mild-moderate non-syndromic sensorineural hearing loss, J. Med. Genet. 38 (2001) 20—25. [13] I. Lerer, M. Sagi, Z. Ben-Neriah, T. Wang, H. Levi, D. Abeliovich, A deletion mutation in GJB6 cooperating with a GJB2 mutation in trans in non-syndromic deafness: a novel founder mutation in Ashkenazi Jews, Hum. Mutat. 18 (2001) 160. [14] I. del Castillo, M. Villamar, M.A. Moreno-Pelayo, F.J. del Castillo, A. Alvarez, D. Telleria, et al. A deletion involving the connexin 30 gene in nonsyndromic hearing impairment, N. Engl. J. Med. 346 (2002) 243—249. [15] N. Pallares-Ruiz, P. Blanchet, M. Mondain, M. Claustres, A.F. Roux, A large deletion including most of GJB6 in recessive non syndromic deafness: a digenic effect? Eur. J. Hum. Genet. 10 (2002) 72—76. [16] B. Teubner, V. Michel, J. Pesch, J. Lautermann, M. CohenSalmon, G. Sohl, et al. Connexin30 (Gjb6)-deficiency causes severe hearing impairment and lack of endocochlear potential, Hum. Mol. Gen. 12 (2003) 13—21. [17] M.H. Kemperman, L.H. Hoefsloot, C.W.R.J. Cremers, Hearing loss and connexin 26, J. R. Soc. Med. 95 (2002) 171—177. [18] L.H. Hoefsloot, M.H. Kemperman, C.W.R.J. Cremers, Van gen naar ziekte; slechthorendheid en connexine 26, Nederlands Tijdschrift Geneeskunde 146 (2002) 259—261. [19] C. Stinckens, H. Kremer, E. van Wijk, L.H. Hoefsloot, P.L.M. Huygen, L. Standaert, et al., Longitudinal phenotypic analysis in patients with connexin 26 (GJB2) (DFNB1) and con-

174

[20]

[21]

[22] [23]

[24]

[25]

R.L.P. Santos et al.

nexin 30 (GJB6) mutations., Ann. Otol. Otorhinolaryngol. 113 (2004) 587—593. J.T. den Dunnen, E. Antonarakis, Nomenclature for the description of human sequence variations, Hum. Genet. 109 (2001) 121—124. M. Mazzoli, G. Van Camp, V. Newton, N. Giarbini, F. Declau, A. Parving, Recommendations for the description of genetic and audiological data for families with nonsyndromic hereditary hearing impairment, in: G. Van Camp, R.J.H. Smith (eds.), Hereditary Hearing Loss Homepage, URL: http:// dnalab-www.uia.ac.be/dnalab/hhh/, Accessed 5 December 2003. X. Liu, L. Xu, Nonsyndromic hearing loss: an analysis of audiograms, Ann. Otol. Rhinol. Laryngol. 103 (1994) 428—433. E.M.R. De Leenheer, P.L.M. Huygen, P.J. Coucke, R.J.C. Admiraal, G. Van Camp, C.W.R.J. Cremers, Longitudinal and cross-sectional phenotype analysis in a new,large Dutch DFNA2/KCNQ4 family, Ann. Otol. Rhinol. Laryngol. 111 (2002) 267—274. D. Botstein, N. Risch, Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease, Nat. Genet. Suppl. 33 (2003) 228—237. A. Pampanos, J. Economides, V. Iliadou, P. Neou, P. Leotsakos, N. Voyiatzis, et al. Prevalence of GJB2 mutations in prelingual deafness in the Greek population, Int. J. Pediatr. Otolaryngol. 65 (2002) 101—108.

[26] F. Gualandi, A. Ravani, A. Berto, A. Sensi, C. Trabanelli, F. Falciano, et al. Exploring the clinical and epidemiological complexity of GJB2-linked deafness, Am. J. Med. Genet. 112 (2002) 38—45. [27] S.A. Wilcox, K. Saunders, A.H. Osborn, A. Arnold, J. Wunderlich, T. Kelly, et al. High frequency hearing loss correlated with mutations in the GJB2 gene, Hum. Genet. 106 (2000) 399—405. [28] A.J. Griffith, A.A. Chowdhry, K. Kurima, L.J. Hood, B. Keats, C.I. Berlin, et al. Autosomal recessive nonsyndromic neurosensory deafness at DFNB1 not associated with the compound-heterozygous GJB2 (connexin 26) genotype M34T/ 167delT, Am. J. Hum. Genet. 67 (2000) 745—749. [29] M.A. Kenna, B.L. Wu, D.A. Cotanche, B.R. Korf, H.L. Rehm, Connexin 26 studies in patients with sensorineural hearing loss, Arch. Otolaryngol. Head Neck Surg. 127 (2001) 1037— 1042. [30] I. del Castillo, M.A. Moreno-Pelayo, F.J. del Castillo, Z. Brownstein, S. Marlin, Q. Adina, et al. Prevalence and evolutionary origins of the del(GJB6-D13S1830) mutation in the DFNB1 locus in hearing-impaired subjects: a multicenter study, Am. J. Hum. Genet. 73 (2003) 1452—1458. [31] M. Cohen-Salmon, T. Ott, V. Michel, J.P. Hardelin, I. Perfettini, M. Eybalin, et al. Targeted ablation of connexin26 in the inner ear epithelial gap junction network causes hearing impairment and cell death, Curr. Biol. 12 (2002) 1106—1111.